Novotarskyi Mykhailo Anatolievich
Education
Graduated from the National Technical University of Ukraine "Kyiv Polytechnic Institute", 1979.
Major: Computer Engineering.
Major: Computer Engineering.
Dissertations
1. Ŕwarded the academic degree of Candidate of Technical Sciences in Computers, complexes, systems and networks. 1988.
Thesis title: "Features of the construction and operation of asynchronous multiprocessor systems with local interaction of processors".
2. Awarded the academic degree of Doctor of Technical Sciences in Mathematical modeling and computational methods. 2010.
Thesis title:"Parallel asynchronous methods and tools for modeling the processes of peristaltic"
Thesis title: "Features of the construction and operation of asynchronous multiprocessor systems with local interaction of processors".
2. Awarded the academic degree of Doctor of Technical Sciences in Mathematical modeling and computational methods. 2010.
Thesis title:"Parallel asynchronous methods and tools for modeling the processes of peristaltic"
Research interests
The first studies where devoted to solving problems of synthesis of homogeneous parallel computing structures, focused on solving problems, which are characterized by a natural parallelism. The procedure for the synthesis of homogeneous asynchronous multiprocessor systems where proposed. A new approach to simulations where created. It uses the special nets for the formal description.
A number of articles were devoted to the problems of effective implementation of asynchronous parallel methods for solving non-stationary boundary problems of mathematical physics on parallel computer systems. In papers of this cycle were also considered the problems of using boundary value problems of mathematical physics as workload of the model.
Studies of structural and functional characteristics of neural networks using different types of neurons were carried out in 1999-2002. The influences of training on self-organization of computational process for solving boundary problems of mathematical physics by numerical methods was analyzed for several architectures of neural networks. The model of a digital neuron, which based on the latest achievements of neurobiology in the field of synaptic interaction, was proposed.
Architecture of two-level discrete cellular neural network based on digital models of neuron for solving partial differential equations by locally-asynchronous multigrid method was designed. According to research some software systems for simulating three-dimensional cellular neural networks, focused on solving boundary value problems that describe the hydrodynamic processes in the areas of complex shape were developed.
During 2003-2011 was developed locally-asynchronous multigrid parallel method for solving equations of mathematical physics, which features a modified principle of asynchronous interaction between parallel processes, the local nature of the exchange of data between computing nodes and using multigrid calculations that can increase the degree of parallelism of calculations and improve efficiency of equipment.
Topology of three-dimensional circulant graph based on association of topological properties of two-dimensional optimal circulant graph and three-dimensional torus, which allowed to reduce the diameter of the graph and its average distance compared to the individual two-dimensional circulant graph and three-dimensional torus was proposed.
New methods of construction of optimal routes in a three-dimensional circulant graphs were developed. Their main features consist in local nature of the decision to choose a route and the possibility of uniform distribution of communication load that allows more reliable delivery of data.
The model of cellular neural network for numerical solution of partial differential equations, which reduces the average time communications spent on the iteration, if we use the three-dimensional circulant graph as a basic structure of relations between the nodal processes was developed.
New tool for the formal description of models for simulating cellular neural networks on parallel computer systems was created. This tool has the properties of process modeling and simulation of discrete events with APRO-nets and includes a scripting language to describe the rules of functioning of the model and data structures.
The formal description of three-dimensional structures of cellular neural networks for solving partial differential equations on parallel computing systems by locally-asynchronous multigrid method was developed. With this description of the model we were able to simplify the presentation of models in the software.
These research results used in the performance of agreements on scientific cooperation:
1. "Development of new methods of reconstruction operations of the digestive canal by mathematical modeling and computational experiment." Agreement of the National Medical University. O.O. Bogomolets and the Institute of Mathematics of NAS of Ukraine:
2. "Principles of development of reconstructive operations on the digestive tube by mathematical modeling and evaluation of their effectiveness." Agreement of the Vinnitsa National Medical University. M.I. Pirogov and the Institute of Mathematics of NAS of Ukraine.
In the framework of cooperation we created methods and tools of mathematical modeling of reconstructive operations on the digestive tract. A simulation of reconstructive operations, enabling the to conduct quantitative comparison of hydrodynamic parameters of the processes after major resection of hollow organs of the digestive tract, and to determine optimal approaches to reduce the risk of short bowel syndrome was realized.
A number of articles were devoted to the problems of effective implementation of asynchronous parallel methods for solving non-stationary boundary problems of mathematical physics on parallel computer systems. In papers of this cycle were also considered the problems of using boundary value problems of mathematical physics as workload of the model.
Studies of structural and functional characteristics of neural networks using different types of neurons were carried out in 1999-2002. The influences of training on self-organization of computational process for solving boundary problems of mathematical physics by numerical methods was analyzed for several architectures of neural networks. The model of a digital neuron, which based on the latest achievements of neurobiology in the field of synaptic interaction, was proposed.
Architecture of two-level discrete cellular neural network based on digital models of neuron for solving partial differential equations by locally-asynchronous multigrid method was designed. According to research some software systems for simulating three-dimensional cellular neural networks, focused on solving boundary value problems that describe the hydrodynamic processes in the areas of complex shape were developed.
During 2003-2011 was developed locally-asynchronous multigrid parallel method for solving equations of mathematical physics, which features a modified principle of asynchronous interaction between parallel processes, the local nature of the exchange of data between computing nodes and using multigrid calculations that can increase the degree of parallelism of calculations and improve efficiency of equipment.
Topology of three-dimensional circulant graph based on association of topological properties of two-dimensional optimal circulant graph and three-dimensional torus, which allowed to reduce the diameter of the graph and its average distance compared to the individual two-dimensional circulant graph and three-dimensional torus was proposed.
New methods of construction of optimal routes in a three-dimensional circulant graphs were developed. Their main features consist in local nature of the decision to choose a route and the possibility of uniform distribution of communication load that allows more reliable delivery of data.
The model of cellular neural network for numerical solution of partial differential equations, which reduces the average time communications spent on the iteration, if we use the three-dimensional circulant graph as a basic structure of relations between the nodal processes was developed.
New tool for the formal description of models for simulating cellular neural networks on parallel computer systems was created. This tool has the properties of process modeling and simulation of discrete events with APRO-nets and includes a scripting language to describe the rules of functioning of the model and data structures.
The formal description of three-dimensional structures of cellular neural networks for solving partial differential equations on parallel computing systems by locally-asynchronous multigrid method was developed. With this description of the model we were able to simplify the presentation of models in the software.
These research results used in the performance of agreements on scientific cooperation:
1. "Development of new methods of reconstruction operations of the digestive canal by mathematical modeling and computational experiment." Agreement of the National Medical University. O.O. Bogomolets and the Institute of Mathematics of NAS of Ukraine:
2. "Principles of development of reconstructive operations on the digestive tube by mathematical modeling and evaluation of their effectiveness." Agreement of the Vinnitsa National Medical University. M.I. Pirogov and the Institute of Mathematics of NAS of Ukraine.
In the framework of cooperation we created methods and tools of mathematical modeling of reconstructive operations on the digestive tract. A simulation of reconstructive operations, enabling the to conduct quantitative comparison of hydrodynamic parameters of the processes after major resection of hollow organs of the digestive tract, and to determine optimal approaches to reduce the risk of short bowel syndrome was realized.
Experience
1. 03.1979-07.1980 — Designer-Trainee third category of the Institute for Superhard Materials, Kyiv.
2. 07.1980-07.1987 — Designer third category of the Institute for Superhard Materials, Kyiv.
3. 07.1987-09.1987 — Senior Engineer of the Institute of Mathematics of NAS, Ukraine.
4. 09.1987-02.1991 — Junior Scientific Researcher of the Institute of Mathematics of NAS, Ukraine.
5. 02.1991-01.2002 — Scientific Researcher of the Institute of Mathematics of NAS, Ukraine.
6. 01.2002-04.2011 — Senior Scientific Researcher of the Institute of Mathematics of NAS, Ukraine.
7. 04.2011-present — Head Scientific Researcher of the Institute of Mathematics of NAS, Ukraine.
2. 07.1980-07.1987 — Designer third category of the Institute for Superhard Materials, Kyiv.
3. 07.1987-09.1987 — Senior Engineer of the Institute of Mathematics of NAS, Ukraine.
4. 09.1987-02.1991 — Junior Scientific Researcher of the Institute of Mathematics of NAS, Ukraine.
5. 02.1991-01.2002 — Scientific Researcher of the Institute of Mathematics of NAS, Ukraine.
6. 01.2002-04.2011 — Senior Scientific Researcher of the Institute of Mathematics of NAS, Ukraine.
7. 04.2011-present — Head Scientific Researcher of the Institute of Mathematics of NAS, Ukraine.